Stress Map after Topology Optimization

Francesco Cataldi_21577

Hi, I'm doing for my Master Thesis a Topology Optimization of a component using solid Tetra elements. As constrints I put volume fraction and the wheighted compliance as an objective. I'm also using 6 loadsteps simultaneously. Is there a way to get the stress map directly on the optimized part or I need to do another static analysis with the new geometry? Is there a control card I can activate to obtain this result?

Tank you

Adriano A. Koga

yes and no.

 

OptiStruct will write an H3D file for each loadcase that you have, containing stress/displacement for the initial and final iteration, by default.

You can load this into HyperView and plot the contour of the final iteration to have an idea of stress levels for your part. You could combine contour and iso view to hide the lower densitiy elements. The issue here, is that it also takes into account the lower desnsity elements as well.

So in order to have a better view of the stress, ideally you can go back to HyperMesh and use OSSMOOTH (FEA Reanalysis) capabilities to get an updated model, with the changes proposed by the TO, and then run again this model. This would give you a better understanding of your new concept.

Francesco Cataldi_21577

Adriano A. Koga_21884 said:

yes and no.

 

OptiStruct will write an H3D file for each loadcase that you have, containing stress/displacement for the initial and final iteration, by default.

You can load this into HyperView and plot the contour of the final iteration to have an idea of stress levels for your part. You could combine contour and iso view to hide the lower densitiy elements. The issue here, is that it also takes into account the lower desnsity elements as well.

So in order to have a better view of the stress, ideally you can go back to HyperMesh and use OSSMOOTH (FEA Reanalysis) capabilities to get an updated model, with the changes proposed by the TO, and then run again this model. This would give you a better understanding of your new concept.

Thank you for your answer. By looking at the iteration 0 stresses I could see they are different from the static analysis I did before the optimization. Is it OK or there is a mistake somewhere?

Adriano A. Koga

Francesco Cataldi_21577 said:

Thank you for your answer. By looking at the iteration 0 stresses I could see they are different from the static analysis I did before the optimization. Is it OK or there is a mistake somewhere?

iteration zero of the optimization is based on the INITIAL value of the design variable, which might be different from the original property value.

You will see at Iter. 0 of your .out file a statement marked with * saying that.

Let`s say shell thickness is 10.0 in the PSHELL.

But when you create your SIZE optimization, you define DESVAR with 8.0 as initial value.

OS will yrun your model using 8.0 instead of the original 10.0, because you have provided a different initial value for the optimization.